A Volume-Preserving Free-Form Deformation Technique for Customising a Face Model to Another Configuration

Abstract

This article presents a volume-preserving freeform deformation technique that can be used to customise a face model generated from magnetic resonance (MR) images into another configuration using only the surface (skin) data. This customisation process is useful when comparing anatomical measurements between datasets that may have undergone a different mode deformation. For example, gravity and other body forces were often neglected in most biomechanical simulations, and as a result, a supine face model generated from MR images is not suitable for analysis of activities performed in upright posture (i.e. expression detection for human computer interaction). To address this problem, the supine model can be fitted to the scanned skin data of an upright posture, in which conventional biomechanical simulations can be applied. Volume-preservation is an important characteristic of soft tissue deformation due to the high water content, and therefore is essential to produce realistic results, especially when only surface information is available. Validation studies presented in this article showed good agreement between the actual deformations and the predictions by the proposed method.